Create Elasticsearch indexes for Inventory search

Description

Current situation or problem:

The existing inventory search does not provide enough flexibility in building the search criteria, does not perform as expected when searches in large collection takes long to return the data. Also, in searches that have more than 1000 records fulfilling the search, the count of the matching records is incorrect.

In scope
Searches against inventory instances, holdings and items:

Instance

Holdings

Items

Keyword search (title, contributor, identifier)

Effective location

Item Status

Effective location

Holdings permanent location

Effective location

Language

Suppress from discovery

Holdings permanent location

Resource type

 

Material type

Format

 

Suppress from discovery

Mode of issuance

 

 

Nature of content

 

 

Staff suppress

 

 

Suppress from discovery

 

 

Date created (from, to)

 

 

Date updated (from, to)

 

 

Source

 

 

Out of scope

  1. Other than inventory records searches

  2. Updates to existing inventory search.

Proposed solution/stories

1. Define mappings that will determine

  • which fields should be treated as full text search fields

  • which fields contain numbers and dates and their format

  • custom rules for dynamically added fields

2. Define approach to text analysis (tokenization and normalization)

Links to additional info

Questions

Priority

Fix versions

Development Team

Falcon

Assignee

Solution Architect

Parent Field Value

None

Parent Status

None

blocks

is defined by

Checklist

hide

TestRail: Results

Activity

Show:

Steve Bischof December 11, 2020 at 6:47 PM

Probably the most obvious use case for Five Colleges is limiting by location in combination with things like identifiers, title, author. With multiple cataloging, acquisition and circulation units, this becomes a critical tool for finding the appropriate instance/holding/item

Magda Zacharska December 11, 2020 at 6:39 PM

The initial implementation of Elasticsearch for inventory does not mean to replace the existing Inventory search but add additional option for the user. and if this is showstopper for your institutions please provide use cases so we can prioritize.

Tod Olson December 10, 2020 at 9:43 PM

Chicago - The issue for us is around needed improvements to search and search results in Inventory, many of which have become tied to Elasticsearch implementation. The problem with search in Inventory is that there are many cases where the available identifying text is not enough distinguish between the results. Without those needed improvements, we will be having some difficult conversations about whether we can support our workflows.

Magda Zacharska December 9, 2020 at 11:05 PM

Hi could you provide some use cases where the current implementation of Inventory search is insufficient for Five Colleges' needs?

Steve Bischof December 9, 2020 at 7:46 PM

Five College Showstopper based the complexity of our locations

Done

Details

Reporter

Front End Estimate

XXL < 30 days

Front End Estimator

Front-End Confidence factor

Low

Back End Estimate

XXL < 30 days

Back End Estimator

Score

16

Rank: 5Colleges (Full Jul 2021)

R1

Rank: Cornell (Full Sum 2021)

R1

Rank: GBV (MVP Sum 2020)

R2

Rank: TAMU (MVP Jan 2021)

R2

Rank: Chicago (MVP Sum 2020)

R1

Rank: U of AL (MVP Oct 2020)

R2

Rank: Lehigh (MVP Summer 2020)

R2

TestRail: Cases

Open TestRail: Cases

TestRail: Runs

Open TestRail: Runs
Created November 10, 2020 at 4:24 PM
Updated January 17, 2022 at 5:11 PM
Resolved April 19, 2021 at 9:46 PM
TestRail: Cases
TestRail: Runs